- GRU2GRU : Project name
- consonant-vowel : Use consonant and vowel to make output
- flip : Flip the input of the models
- no_dropout : Not used
dropout
layer fordecoder
- LSTM : Use
LSTM
forencoder
GRU2GRU_consonant-vowel_LSTM_flip
has the best accuracy with 94%
(You have to wait until it boots) https://english2korean.herokuapp.com/
If you want to see the graphs, check the results
or use :
$ tensorboard --logdir=./log